ComfyUI_VLM_nodes icon indicating copy to clipboard operation
ComfyUI_VLM_nodes copied to clipboard

Published 20 hours ago • verified publisher icon gokayfem

llava quant Not staying loaded

Open pinballelectronica opened this issue 10 months ago • 0 comments

Hey. LOVE these nodes. Llava Optional Memory Free Advanced seems to not actually cache the model between runs. Unload is false. Control after gen, fixed. Am I missing something? running 13b vicuna q6km quant with the associated projector. Thank you.

pinballelectronica avatar Dec 21 '24 14:12 pinballelectronica